The Joint Adaptive Kalman Filter (JAKF) for Vehicle Motion State Estimation

Abstract

This paper proposes a multi-sensory Joint Adaptive Kalman Filter (JAKF) through extending innovation-based adaptive estimation (IAE) to estimate the motion state of the moving vehicles ahead. JAKF views Lidar and Radar data as the source of the local filters, which aims to adaptively adjust the measurement noise variance-covariance (V-C) matrix ‘R’ and the system noise V-C matrix ‘Q’. Then, the global filter uses R to calculate the information allocation factor ‘β’ for data fusion. Finally, the global filter completes optimal data fusion and feeds back to the local filters to improve the measurement accuracy of the local filters. Extensive simulation and experimental results show that the JAKF has better adaptive ability and fault tolerance. JAKF enables one to bridge the gap of the accuracy difference of various sensors to improve the integral filtering effectivity. If any sensor breaks down, the filtered results of JAKF still can maintain a stable convergence rate. Moreover, the JAKF outperforms the conventional Kalman filter (CKF) and the innovation-based adaptive Kalman filter (IAKF) with respect to the accuracy of displacement, velocity, and acceleration, respectively.

Aircraft tracking based on fully conventional network and Kalman filter

“…In addition, due to the fact that the KF can predict the future state of the object according to the state model and the observation vector, a number of relevant works have been done to evaluate the target's trajectory, as in [31, 32]. Beyond that, there are many practical problems that are not completely linear, so extended KF (EKF) [33] is presented to fit the non‐linear system, EKF carries out first‐order Taylor expansion of the non‐linear part and employs Jacobian matrix to replace the constant on KF.…”

A Novel Method for the Dynamic Coefficients Identification of Journal Bearings Using Kalman Filter

“…This method has been used successfully in the estimation of the critical parameters of the system, such as force [19][20][21][22], structural damage diagnosis [23], inverse heat conduction [24], pore water electrical conductivity [25], and mobile-robot attitude [26] and dynamic state [27][28][29]. Additionally, compared with other algorithms, such as dual Kalman filter [30], join Kalman filter [31], and even recursive least squares (RLS) [32], Kalman filtering is not only easier to achieve for estimating the main parameters in the discrete-time dynamic system, but also can save computing time.…”

A Model-Based Method for Estimating the Attitude of Underground Articulated Vehicles

“…Therefore, automated UAVs (AUAVs) have first priority in underground mines. However, the automation of their precise control and navigation requires related information on the target vehicle [4][5][6], which includes the attitude, velocities, and even accelerations relative to different directions [7].…”